Refractory Celiac Disease: What Changes When the Cells Themselves Are the Diagnosis
Two patients, both refractory to a strict gluten-free diet, both diagnosed by the same flow cytometry panel — but one result comes back clean and the other comes back aberrant, and that single laboratory line changes what kind of disease the team is actually treating.
Frank D., a 61-year-old retired machinist, spent thirty-nine years running a lathe shop floor before his knees forced him into retirement two years ago — around the same time abdominal cramping and diarrhea that had been building for months finally sent him to a gastroenterologist. His celiac serology and biopsy were unambiguous at diagnosis, and he adopted a gluten-free diet within the week, tracked with the same precision he once brought to machine tolerances. Ten months in, with dietitian-confirmed adherence and no dietary lapse either of them can identify, his diarrhea has actually worsened, and a repeat biopsy still shows Marsh IIIa villous blunting.
Flow cytometry on that repeat biopsy found a normal, polyclonal intraepithelial lymphocyte population — the finding that defines his disease as refractory celiac Type I rather than Type II, and with it, a genuinely lower risk trajectory: cohort data (Al-Toma et al.; Malamut et al.) put Type I's own progression risk to lymphoma in the low single digits, an order of magnitude below Type II's. His nutritional numbers say something similar about pace. Six pounds over ten months, with an albumin still at 3.4, is a slow drift rather than the steep loss that would argue for front-loading immunosuppression — but ten months of dietitian-confirmed adherence producing no histologic improvement whatsoever is not a picture that can be left open-ended either, which makes the reassessment date, rather than the drug, the thing actually worth arguing about. Open-capsule budesonide is the one agent with response data generated in refractory Type I itself rather than borrowed from inflammatory bowel disease (Brar et al.), and induction-dose budesonide is not a drug anyone continues indefinitely. The team's task with Frank is therefore the more familiar induction-versus-maintenance question — which drug gets him into remission, and what holds it once he's there — not the oncologic surveillance question his case-pair partner's flow cytometry result is about to raise.
GI clinic — Frank D.
This is a straightforward induction decision to me. Brar and colleagues generated response data for open-capsule budesonide in refractory Type I itself, not borrowed from inflammatory bowel disease, and Frank’s polyclonal flow cytometry means there’s no oncologic urgency pushing us toward anything heavier. Start there, reassess in a defined window.
I want to push on “defined window,” because I don’t think it is defined yet. Six pounds over ten months with an albumin at 3.4 is genuinely reassuring — I’d say so plainly, and it’s why I’m not arguing for immunosuppression here the way I am for Judith. But a shallow curve measured once is not the same as a stable one. We have a single albumin, not a trend, and the reason we’re here at all is that ten months of confirmed adherence produced no histologic movement. “Low progression risk” and “slow enough that we can reassess whenever” are not the same claim, and only the first one is what his flow cytometry actually supports.
Both of you are still stopping at induction, and that’s the part I’d contest. Induction-dose budesonide is not a drug we continue indefinitely — so a plan that names no successor isn’t the conservative option, it’s an unfinished one, and it quietly commits him to having this exact conversation again the moment the drug works. Name azathioprine in the chart today as the maintenance agent, and set the reassessment at six weeks with a repeat albumin, so the hematologist gets his second data point on the same visit that tells us whether budesonide is working at all.
Agreed: budesonide starts today with a six-week reassessment and azathioprine named as the maintenance plan in advance.
Judith A., a 66-year-old retired librarian, had run the same small-town branch library for over two decades before retiring three years ago, and still volunteers there most Tuesdays shelving returns. Her celiac disease was diagnosed eight years ago, well controlled on a gluten-free diet for the first seven, until a slow return of diarrhea and an eighteen-pound weight loss over the past year brought her back for re-evaluation. Repeat biopsy confirmed persistent villous atrophy despite her dietitian-verified adherence — the same refractory picture Frank’s biopsy showed.
Her flow cytometry did not come back the same way. It found an aberrant, immunophenotypically abnormal intraepithelial lymphocyte population — the defining feature of refractory celiac disease Type II — and that single result reframes everything downstream of it. Type II carries a quantified progression risk to enteropathy-associated T-cell lymphoma that cohort series (Al-Toma et al.; Malamut et al.) place at roughly a third to half of patients within five years, an order of magnitude above Type I's own risk. Her own numbers do not sit neutrally inside that range. The prognostic model Rubio-Tapia and colleagues built for refractory disease scores age at or above 65, hypoalbuminemia, anemia and aberrant intraepithelial lymphocytes; at 66, with eighteen pounds gone in a year and an aberrant clone on flow cytometry, she carries several of them before anyone has ordered a staging scan. That is the difference between Frank's six pounds over ten months and hers — the same phrase, weight loss, describing a slow drift in one case and a scored prognostic factor in the other. The evidence for treating the clonal population directly — cladribine, aimed at cytoreducing the aberrant cells themselves rather than simply controlling symptoms — comes from real but genuinely smaller and older series (Al-Toma et al.) than the Type I budesonide data her case-pair partner's plan rests on, and nothing about her current imaging has yet confirmed whether this is truly an isolated clonal finding or early, not-yet-visible lymphoma already present underneath it.
GI/hematology joint clinic — Judith A.
This changes what we’re actually treating. An aberrant clonal population isn’t a more severe version of Frank’s refractory disease — it’s a premalignant clonal finding with a real, quantified progression risk to lymphoma, put by Al-Toma and Malamut at roughly a third to half of patients within five years — and she is 66 with eighteen pounds gone in a year, which already places her on the wrong side of Rubio-Tapia’s own prognostic factors before we have staged anything. Symptom control alone is not an adequate endpoint here the way it was for Frank. We need to be treating the clonal burden itself, and I want cladribine on the table now, not after a budesonide trial that was never designed with this population in mind.
I don’t disagree that this is a different disease in kind, not just severity — but I want to be honest about how much thinner the cladribine evidence actually is. It’s real — Al-Toma’s cladribine series is a genuine result — but it comes from smaller, older numbers than the Type I budesonide data, and cytoreductive therapy is not a benign intervention to reach for on a comparatively thin evidence base without more information first.
I’m not arguing against treating the clonal population — I’m arguing that "the evidence supports treating this differently" and "the evidence supports this specific drug, at this specific time" are two different claims, and only the first one is as solid as it sounds.
The piece I think is actually missing from both positions is staging. Nobody has confirmed yet whether this is a genuinely isolated aberrant clonal population or early overt EATL that just hasn’t been imaged. Starting cladribine, a drug whose entire purpose is different from anything in Frank’s regimen, on an assumption about disease extent that hasn’t been tested, risks treating the wrong stage of the wrong problem.
I’d sequence this: PET-CT and device enteroscopy staging first, on an expedited timeline given her weight loss, then a cladribine decision made with actual extent-of-disease information in hand — not instead of treating the clonal population seriously, but so that when we do treat it, we know what we’re actually treating.
Agreed: expedited PET-CT and device enteroscopy staging, with hematology-oncology and gastroenterology reviewing results jointly before any cytoreductive therapy decision.
Not agreed: whether cladribine should start immediately if staging confirms isolated clonal disease with no overt lymphoma, or whether even that finding should prompt a slower, more heavily monitored start given the size of the evidence base. The oncologist favors starting promptly once staging clears; the gastroenterologist wants a lower starting dose and closer interval monitoring than the oncologist considers necessary.